CoChat
Sign Up
Cited by 25Teaching and Learning ProgrammingOnline and Blended LearningInnovative Teaching Methods

The effects of online peer-facilitated learning and distributed pair programming on students’ learning

Open access
Tsai, Chia-Wen, Lin, Michael Yu-Ching, Cheng, Yih-Ping, Lee, Lan-Yu, Chyr, Wen-Li, Lin, Chih-Hsien, Lin, Jian-Wei, Tsai, Meng-Chuan·2023·Computers & Education
DOI 10.1016/j.compedu.2023.104849PMC PMC10232937
Open this paper in CoChat — free accountOpen PDF

Abstract

The COVID-19 pandemic has forced universities and schools around the world to adopt online learning. Teachers may wonder if their students can attain satisfactory learning performance in an online learning environment without teachers' on-the-spot attention. In order to develop students' skills in programming, promote their enjoyment of learning and intention to learn to program, the researchers integrated two innovative teaching approaches, using online peer-facilitated learning and distributed pair programming, and investigated the effects of these on students' online learning performance. This study conducted an experiment that included 128 undergraduates from four class sections of Department of Finance. Thus, the experimental design in this research was a 2 (Peer-facilitated learning vs. non-peer-facilitated learning) × 2 (Distributed pair programming vs. non-distributed pair programming) factorial pretest/post-test design. The participants in this research mainly consisted of four classes of students from a non-computer or information department who took a compulsory course on programming design.’ Both quantitative and qualitative data were collected in this study. According to the results, the peer-facilitated learning group exhibited significantly better development of programming skills, enjoyment of learning, and intention to learn, than the non-peer-facilitated learning group. However, expected effects of enhancing the learning of the students in this study who received the distributed pair programming were not found. The design of online pedagogy can be a reference for online educators. The implications of applying online peer-facilitated learning and distributed pair programming to support students’ learning and the design of online programming courses are discussed.

AI snapshot

Written from the paper’s full text.

TL;DR

This study tested online peer-facilitated learning and distributed pair programming in a 2×2 quasi-experiment with 128 non-computing undergraduates in an online VBA programming course. Peer-facilitated learning improved programming skills, enjoyment, and intention to learn, while distributed pair programming did not show expected benefits and was associated with lower programming skills.

Key findings

  • Peer-facilitated learning group (G1+G2, n=66) significantly outperformed the non-peer-facilitated group (G3+G4, n=62) on programming skills (M=78.379 vs 72.275, p=.001), enjoyment of learning (M=3.283 vs 2.952, p=.029), and intention to learn (M=4.652 vs 4.075, p=.018).
  • Distributed pair programming group (G1+G3, n=60) did not outperform the non-distributed group (G2+G4, n=68); programming skills were significantly lower in the distributed pair group (M=71.634 vs 78.765, p=.000), with no significant differences for enjoyment (p=.273) or intention (p=.162).
  • The combined intervention group G1 did not significantly outperform other groups. G3 (distributed pair programming only) had significantly lower programming skills than G2 (peer-facilitated only, p=.000) and G4 (control, p=.020).
  • Qualitative interviews were mixed: some students valued peer collaboration and immediate paired assistance, while others reported that paired groups followed the teacher without discussion, had conflicts, or paid little attention to the paired group.

Methods

  • Design: 2 (peer-facilitated vs. non-peer-facilitated) × 2 (distributed pair programming vs. non-distributed pair programming) factorial pretest/post-test quasi-experiment. Participants were 128 first-year undergraduates from four class sections of a compulsory “Programming Design” course in a Taiwanese university; mean age 19.25. Group sizes: G1 n=25, G2 n=41, G3 n=35, G4 n=27.
  • Course/interventions: 17-week online course using Visual Basic for Applications (VBA). Peer-facilitated learning used teams of six and a seven-phase peer coaching process; distributed pair programming used pairs working remotely via Microsoft Teams, TeamViewer, or LINE/Facebook outside class. The control group received traditional lectures without either intervention.
  • Measures/analysis: programming skills were graded from student VBA projects using six DeLone & McLean IS success dimensions; enjoyment of learning used three AEQ items on a 5-point Likert scale; intention to learn used three items on a 7-point Likert scale. Analysis included CFA, Levene’s test, independent-sample t-tests, one-way ANOVA, and content analysis of semi-structured interviews with 20 students (5 per group).

Limitations

  • The paper states possible Hawthorne effect, disturbance from the COVID-19 outbreak and students’ health situations, and threats from the quasi-experimental design that may limit firm conclusions.
  • Evident from the text: groups were intact class sections rather than randomly assigned; some groups were small (e.g., G1 n=25, G4 n=27); enjoyment and intention were self-reported; and distributed pair programming implementation issues (students imitating the teacher, conflicts, limited discussion) may have affected the results.

Model-written summary. Check it against the paper before relying on it.

Integrity signals

Registry

Crossref · 10.1016/j.compedu.2023.104849

Cites

114 of 114 references verified0 citation errors · 0 not checkable

Statistics

Nothing to recompute.

Claims the paper makes

1 of 25 cited sources checked · 1 supported · 0 flagged

  • Premisecitation
    § 2.4. Enjoyment of learning · p. 5

    Learners in pair programming enjoy programming collaboratively.

    “In addition, it is revealed that learners in pair programming enjoy programming collaboratively (Bowm an et al., 2020; Faja, 2011).”
    Quote in the paperSubstantial claim1 of 2 sources checked
    • cites Bowm an et al., 2020

      Couldn’t be matched to a paper in Crossref, so it can’t be checked.

    • cites Faja, 2011 source
  • Findingown data
    § A B S T R A C T · p. 2

    The expected effects of distributed pair programming on enhancing students' learning were not found in this study.

    “However, expected effects of enhancing the learning of the students in this study who received the distributed pair programming were not found.”
    Quote in the paperSubstantial claim
  • Findingown data
    § 4.3. The effect of distributed pair programming · p. 10

    There was no significant difference in intention to learn between the distributed pair programming group and the non-distributed pair programming group (p>.05).

    “Table 3 reveals no significant difference ( p > .05) in students ' enjoyment of learning and intention to learn.”
    Quote in the paperSubstantial claim
  • Findingown data
    § 4.3. The effect of distributed pair programming · p. 10

    There was no significant difference in enjoyment of learning between the distributed pair programming group and the non-distributed pair programming group (p>.05).

    “Table 3 reveals no significant difference ( p > .05) in students ' enjoyment of learning and intention to learn.”
    Quote in the paperSubstantial claim
  • Findingown data
    § 4.2. The effect of peer-facilitated learning · p. 10

    In the posttest, the peer-facilitated learning group (n=66) had a mean intention to learn score of 4.652, significantly higher than the non-peer-facilitated learning group (n=62, mean=4.075), t=2.386, p=.018.

    “Intention to learn | 66 | 4.652 | 1.309 | 7.000 | 1.000 | 62 | 4.075 | 1.422 | 7.000 | 1.000 | 2.386 | 126 | .018*”
    Quote in the paperSubstantial claim
  • May combine two claims. 17
  • The quote only partly states this claim. 1
  • Classified as probably not a substantial claim. 17

  • May combine two claims. The quote only partly states this claim.

    Findingown data
    § 4.1. Validity and reliability · p. 9

    Cronbach's alpha values for enjoyment of learning and intention to learn are 0.890 and 0.917, both greater than 0.7.

    “In addition, Cronbach ' s α values are 0.890 and 0.917, both greater than 0.7.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingown data
    § 4.1. Validity and reliability · p. 9

    The composite reliability values for enjoyment of learning and intention to learn are 0.909 and 0.958, both greater than 0.7.

    “The values of composite reliability for both the variables are greater than 0.7, at 0.909 and 0.958.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingown data
    § 4.2. The effect of peer-facilitated learning · p. 10

    In the posttest, the peer-facilitated learning group (n=66) had a mean programming skills score of 78.379, significantly higher than the non-peer-facilitated learning group (n=62, mean=72.275), t=3.433, p=.001.

    “Programming skills | 66 | 78.379 | 9.635 | 98.000 | 60.000 | 62 | 72.275 | 10.480 | 90.000 | 30.000 | 3.433 | 126 | .001*”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingown data
    § 4.4. The combined effect of peer-facilitated learning and distributed pair programming · p. 10

    The distributed pair programming group (n=60) had a mean programming skills score of 71.634, significantly lower than the non-distributed pair programming group (n=68, mean=78.765), t=4.074, p=.000.

    “Programming skills | 60 | 71.634 | 10.892 | 88.000 | 30.000 | 68 | 78.765 | 8.898 | 98.000 | 66.000 | 4.074 | 126 | .000*”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingown data
    § 4.1. Validity and reliability · p. 9

    All observation items had loading values between 0.83 and 0.97, greater than 0.7.

    “The results found that all observation items had a loading value between 0.83 and 0.97 and were greater than 0.7.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingown data
    § 4.1. Validity and reliability · p. 9

    The AVE values for enjoyment of learning and intention to learn are 0.770 and 0.884, both greater than 0.5.

    “The AVE values for enjoyment of learning and intention to learn are 0.770 and 0.884 respectively, both greater than 0.5.”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Findingown data
    § 4.2. The effect of peer-facilitated learning · p. 10

    In the posttest, the peer-facilitated learning group (n=66) had a mean enjoyment of learning score of 3.283, significantly higher than the non-peer-facilitated learning group (n=62, mean=2.952), t=2.210, p=.029.

    “Enjoyment of Learning | 66 | 3.283 | .794 | 5.000 | 1.330 | 62 | 2.952 | .901 | 5.000 | 1.000 | 2.210 | 126 | .029*”
    Quote in the paperSubstantial claim

  • Classified as probably not a substantial claim. May combine two claims.

    Premisecitation
    § 2.5. Intention to learn · p. 5

    Learners who receive peer-facilitated learning are motivated and engage more in the learning material.

    “In addition, it is reported that learners who receive peer-facilitated learning are motivated and engage more in the learning material, as they find it is easier to ask questions while they are being taught or responded to by students of similar age (Mkonto, 2018).”
    Quote in the paperSubstantial claim0 of 1 sources checked
    • cites Mkonto, 2018 source

  • Classified as probably not a substantial claim.

    Premisecitation
    § 2.3. Students ' programming skills · p. 5

    There is a lack of research on how to teach programming skills in a practical way that raises students' interest.

    “In addition, it is indicated that there is a lack of research on how to teach programming skills in a practical way that raises students ' interest (Seraj et al., 2019).”
    Quote in the paperSubstantial claim0 of 1 sources checked
    • cites Seraj et al., 2019 source

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Premisecitation
    § 2.3. Students ' programming skills · p. 4

    The need for programming skills and information technology will rapidly rise and may grow as much as 90% between 2016 and 2030.

    “Recent research has indicated that the need for programming skills and information technology will rapidly rise and may grow as much as 90% between 2016 and 2030 (Bughin et al., 2018).”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites Bughin et al., 2018

      Couldn’t be matched to a paper in Crossref, so it can’t be checked.

  • Probably not a substantial claim: may not pass S2 (checkable), S3 (load-bearing).

    Premisecitation
    § 6. Conclusions · p. 13

    It is essential for teachers to develop and adopt appropriate pedagogies to improve students' learning in order to adapt to online learning environments.

    “In order to adapt to online learning environments, it is essential for teachers to develop and adopt appropriate pedagogies to improve students ' learning (Tsai, Shen, & Chiang, 2020).”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites Tsai, Shen, & Chiang, 2020 source

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Study design and conductnone given
    § 5.3. Potential limitations of this research · p. 12

    The experiment environment was disturbed by the COVID-19 outbreak and students' health situations, which might potentially influence the validity of the results.

    “In addition, the experiment environment was disturbed by the COVID-19 outbreak and students ' health situations; these factors might potentially influence the validity of the results.”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Study design and conductown data
    § 3.3.3. Control group (G4) intervention · p. 8

    Students in the control group received the same class hours, learning materials and practice time, and used the same course website as the experimental groups, without the interventions.

    “In this research, students in G4 (the control group) received the same class hours, learning materials and practice time, and used the same course website as those in G1, G2, and G3, however, without the interventions of peer-facilitated learning or distributed pair programming.”
    Quote in the paperSubstantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

    Premisecitation
    § 5.2. The effect of distributed pair programming · p. 12

    Programming is a specialized subject with a high dropout and failure rate.

    “Finally, programming is a specialized subject with a high dropout and failure rate.”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites Dawson et al., 2018 source

  • May combine two claims.

    Interpretationown data
    § 5.1. The effect of peer-facilitated learning · p. 12

    Students who learn in a peer-facilitated learning environment may be more confident in comprehending course content and enjoy their learning.

    “That means students who learn in a peer-facilitated learning environment may be more confident in comprehending course content and enjoy their learning, even in a programming course with historically high drop-out and failure rates.”
    Quote in the paperSubstantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Study design and conductnone given
    § 3.2. Course setting · p. 6

    The course content and course objective were designed and evaluated by a committee of eight teachers with technical and managerial specializations to ensure the course content meets students' needs in the workplace.

    “The course content and course objective were designed and evaluated by a committee of eight teachers with both technical and managerial specializations, to make sure the course content meets students ' needs in the workplace.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Study design and conductcitation
    § 3.4.3. Intention to learn · p. 8

    Intention to learn was measured using scales from Darban et al. (2016) and Davis, Bagozzi, & Warshaw (1989).

    “The researchers in this study adopted scales from Darban et al. (2016) and Davis, Bagozzi, & Warshaw (1989) for behavioral intention to assess their intention to learn programming.”
    Quote in the paperNot a substantial claim0 of 2 sources checked
    • cites Darban et al., 2016 source
    • cites Davis, Bagozzi, & Warshaw, 1989 source

  • May combine two claims.

    Study design and conductown data
    § 3.4.4. Data analysis · p. 8

    The equality of variance was robust and the independent sample t-test and one-way ANOVA could be used.

    “The results indicated that the equality of variance was robust and the independent sample t -test and one-way ANOVA could be used (Huang, Silitonga, & Wu, 2022).”
    Quote in the paperSubstantial claim0 of 1 sources checked
    • cites Huang, Silitonga, & Wu, 2022 source

  • Probably not a substantial claim: may not pass S2 (checkable).

    Interpretationnone given
    § 6. Conclusions · p. 13

    The design of the online peer-facilitated learning instructional method and its effects in an online programming course can provide a reference for teachers to facilitate students' collaboration and improve students' problem-solving skills and learning outcomes.

    “Firstly, the design of online peer-facilitated learning instructional method and its effects in an online pro­ gramming course can provide a reference for teachers to facilitate students ' collaboration and improve students ' problem-solving skills and learning outcomes.”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Interpretationreasoning
    § 4.2. The effect of peer-facilitated learning · p. 9

    When students learn via peer interaction, it may build more profound connections and help students achieve positive performance in programming skills, and better development of enjoyment of learning and intention to learn.

    “When students learn via peer interaction, it may build more profound connections and help students ach­ ieve positive performance in programming skills, and better development of enjoyment of learning and intention to learn.”
    Quote in the paperSubstantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Interpretationreasoning
    § 5.2. The effect of distributed pair programming · p. 12

    These potential factors may have contributed to the ineffectiveness of distributed pair programming found in this study.

    “These potential factors may have contributed to the ineffectiveness of distributed pair programming found in this study.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Study design and conductcitation
    § 5.3. Potential limitations of this research · p. 12

    Students knowing they were being studied might cause the Hawthorne effect.

    “For example, students knowing they were being studied might cause the Hawthorne effect (Chiesa & Hobbs, 2008).”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites Chiesa & Hobbs, 2008 source

  • Classified as probably not a substantial claim.

    Interpretationown data
    § 6. Conclusions · p. 13

    Based on students' interview feedback, the distributed pair programming approach may be effective if further improvements and adjustments are made.

    “Based on students ' feedback in the interview, the distributed pair programming approach may be effective with further improvements and adjustments.”
    Quote in the paperSubstantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Study design and conductcitation
    § 3.4.1. Programming skills · p. 8

    The teacher and researchers in this study blindly graded the programs or systems employing the six success dimensions of system quality, information quality, service quality, usage, user satisfaction, and net benefits.

    “the teacher and researchers in this study blindly graded the programs or systems employing the six success dimensions of system quality, information quality, service quality, usage, user satisfaction, and net benefits.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Study design and conductcitation
    § 3.4.4. Data analysis · p. 9

    A researcher and the involved teacher coded the students' comments.

    “A researcher and the involved teacher in this programming course were invited to code the students ' comments (Chen et al., 2020).”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites Chen et al., 2020 source

  • May combine two claims.

    Premisecitation
    § 5.2. The effect of distributed pair programming · p. 12

    Many students in Taiwan are not willing to learn on their own and are not academically responsible, as they are used to didactic education.

    “Second, many students in Taiwan are not willing to learn on their own and are not academically responsible, as they are used to didactic education (Tsai, 2014; Cheng et al., 2022).”
    Quote in the paperNot a substantial claim0 of 2 sources checked
    • cites Tsai, 2014 source
    • cites Cheng et al., 2022 source

  • May combine two claims.

    Study design and conductown data
    § 4.1. Validity and reliability · p. 9

    The reliability and validity of enjoyment of learning and intention to learn are acceptable.

    “Therefore, the reliability and validity of enjoyment of learning and intention to learn are acceptable, since all the values shown in Table 1 met the standard levels (Huang et al., 2012).”
    Quote in the paperSubstantial claim0 of 1 sources checked
    • cites Huang et al., 2012 source

  • Classified as probably not a substantial claim.

    Study design and conductcitation
    § 3.4.2. Enjoyment of learning · p. 8

    Enjoyment of learning was measured using three items from the Academic Emotions Questionnaire (AEQ) adapted to the programming course.

    “three items from the Academic Emotions Questionnaire (AEQ) (Pekrun et al., 2005; Pekrun, 2006) were slightly adapted to fit the scope of this programming course”
    Quote in the paperSubstantial claim0 of 2 sources checked
    • cites Pekrun et al., 2005

      Couldn’t be matched to a paper in Crossref, so it can’t be checked.

    • cites Pekrun, 2006 source

  • May combine two claims.

    Study design and conductown data
    § 3.4.4. Data analysis · p. 8

    The assumption of homogeneity of variances was met for programming skills (F=1.620, p=.188), enjoyment of learning (F=1.547, p=.206), and intention to learn (F=0.830, p=.480).

    “The results reported homogeneity for programming skills ( F = 1.620, p = .188), enjoyment of learning ( F = 1.547, p = .206), and intention to learn ( F = 0.830, p = .480).”
    Quote in the paperSubstantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

    Premisecitation
    § 5.1. The effect of peer-facilitated learning · p. 12

    Prior studies show that peer-facilitated learning could enhance students' learning outcomes and lead to positive learning attitudes.

    “This result is similar to existing studies showing that peer-facilitated learning could enhance students ' learning outcomes (Karademir, Erdo ˘ gdu & G ¨ okçearslan, 2017; Mahdi, 2006; Mkonto, 2018), and lead to students ' positive learning attitudes (Brown & Rode, 2018; Won & Choi, 2017).”
    Quote in the paperNot a substantial claim0 of 3 sources checked
    • cites Karademir et al., 2017

      Couldn’t be matched to a paper in Crossref, so it can’t be checked.

    • cites Mahdi, 2006 source
    • cites Mkonto, 2018 source

  • Classified as probably not a substantial claim.

    Study design and conductnone given
    § 3.1. Participants · p. 6

    The teacher did not directly interfere with students' interaction and collaboration.

    “The involved teacher adopted and implemented the peer-facilitated learning intervention and distributed pair programming intervention in this study and did not directly interfere with students ' interaction and collaboration.”
    Quote in the paperSubstantial claim
31 of 31

  • The paper relies on this without support.

    Very few studies have explored the effects of peer-facilitated learning in online learning environments.

    “Although the positive effects of peer-facilitated learning have been indicated in various science courses (Lowell et al., 2022), very few studies have explored its effects in online learning environments.”

  • The paper relies on this without support.

    Very few studies on programming education combine and integrate peer-facilitated learning and distributed pair programming in an online course.

    “Up to now, there are very few studies available on programming education that combine and integrate peer-facilitated learning and distributed pair program­ ming in an online course.”
2 of 2

Self-check

Percentages, sums and reported differences re-computed from the paper’s own numbers.

Each contradiction between two passages listed here was confirmed by an independent check.

  • Peer-facilitated group sizes do not fit six-member teams

    Minor

    The two peer-facilitated groups have 25 and 41 students. Neither total can be divided into teams each containing six members, and the text gives no exception for smaller teams.

1 of 1

Cites

Flagged first, then by citations

15 of the 129 references couldn’t be matched to a known paper.

Research works printed but not linked: 3

  1. Hou, 2019, A framework of online judge systems for assessing programming skills, International Journal of Design, Analysis & Tools for Integrated Circuits & Systems, 8, 45
  2. Karademir, 2017, Two birds with one stone: Enhancing technology perception with peer interaction using web 2.0, Journal of Education and Future, 11, 111
  3. Syam, 2022, Online learning in higher education: Analysis during the pandemic Covid-19, Jurnal Mantik, 5, 2256

Cited by

Sorted by citations

Related

Authors
Tsai, C.-W., Lin, M. Y.-C., Cheng, Y.-P., Lee, L.-Y., Chyr, W.-L., Lin, C.-H., Lin, J.-W., & Tsai, M.-C.
Affiliations
  • Ming Chuan University (TW)
  • Dayeh University (TW)
Corresponding author
Tsai, Chia-Wen
ORCID
Venue
Computers & Education
Publisher
Elsevier BV
Published
2023-10
Volume
203
Pages
104849
ISSN
0360-1315 (ISSN-L)
DOI
10.1016/j.compedu.2023.104849
PMC
PMC10232937
OpenAlex
W4379013911
Semantic Scholar
ab36b00bd9c6ede94f5c7fc26d71ec3aca9eaec6
Cited by
25
References
129
Citations by year
2026: 3 · 2025: 11 · 2024: 7 · 2023: 4
FWCI
5.80
Type
journal article
Language
en
Open access
Licence
Locations
  • Computers & Education (publishedVersion)
  • PubMed (publishedVersion)
  • PubMed Central (submittedVersion)
Funders
National Science and Technology Council (110-2628-H-130-001-)
Topics
  • Physical Sciences › Computer Science › Computer Science Applications › Teaching and Learning Programming
  • Social Sciences › Education › Online and Blended Learning
  • Social Sciences › Education › Innovative Teaching Methods
Keywords
distributed pair programming, online programming education, programming skills, joyful learning, intention to learn, non-computer majors, COVID-19 remote teaching

About these checks

Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.

Claims read by deepseek-v4.1-flash under definition version 3 (extract-v3) on 2026-10-12.

Classified by jev-1.13-20260917 (gold-guideline-v0.2) on 2026-10-12.

Cited sources are checked by gpt-6-sol.

Statistics and arithmetic are recomputed, not judged.

Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.

What counts as a claim ›How we measure confidence ›How we check this ›

Are you an author, or do you see a problem with this page? Request a review